To determine whether the third-trimester maternal serum concentration of paraxanthine, caffeine's primary metabolite, is associated with delivery of a small-for-gestational age infant (birth weight less than the 10th percentile for gestational age, gender, and ethnicity) and whether this association differs by smoking, the authors studied 2,515 women who participated in the Collaborative Perinatal Project from 1959 to 1966. The women provided a third-trimester serum sample and had been controls for a nested case-control study of spontaneous abortion. The mean serum paraxanthine concentration was greater in women who gave birth to small-for-gestational age infants (754 ng/ml) than to appropriately grown infants (653 ng/ml, p = 0.02). However, the linear trend for increasing serum paraxanthine concentration to be associated with increasing risk of small-for-gestational age birth was confined to women who also smoked (p = 0.03). There was no association between paraxanthine and fetal growth in nonsmokers (p = 0.48). Adjustment for maternal age, pre-pregnant weight, education, parity, ethnicity, and the number of cigarettes smoked per day did not alter the results substantially, although the p value for trend among smokers increased to 0.07. The authors conclude that maternal third-trimester serum paraxanthine concentration, which reflects caffeine consumption, was associated with a higher risk of reduced fetal growth, particularly among women who smoked.